IDs for events?

This is a summary of something we talked about last meeting - see

http://ilrt.org/discovery/chatlogs/rdfig/2003-07-30.html#T16-21-19

Basically, I tried writing a hacky RDF events generator, and came
across an issue: it would be useful for events to have identifiers. This
came from trying to add myself as an attendee to an existing event found
through a search. Other people have also suggested this.

Some possible solutions discussed:


* Using owl:inverseFunctionalProperty and a URI

16:24:28 <mortenf> what about owl:InverseFunctionalProperties
(foaf-style)?
16:24:47 <libby> that's ok, but we do need to somehow generate a
well-known uri
16:25:43 <libby> uri would be better cos owl:Ifp disnt supposedd to work
with data

16:27:58 <chaalsNCE> get a web service that will return a unique URI
(because it keeps a counter and never gives the same thing twice...)


* Using a UUID (and an inverseFunctionalProperty?)

16:29:12 <ndw> if you're going to make up ids, uuid works really well.
no web service required and statistically certain to be unique
16:29:43 <libby> [[
16:29:43 <libby> 4.8.4.7 Unique Identifier
16:29:44 <libby> Property Name: UID
16:29:45 <libby> Purpose: This property defines the persistent, globally
unique
16:29:47 <libby> identifier for the calendar componen
16:29:49 <libby> ]]
16:30:08 <libby> [[
16:30:09 <libby> identifier is RECOMMENDED to be the identical syntax to  the
16:30:09 <libby> [RFC 822] addr-spec.
16:30:10 <libby> ]]
16:30:33 <libby> that's a standard for arpa text messages
16:30:35 <mortenf> which is id@host, right?
16:30:49 <ndw> that's an ... interesting choice
16:31:13 <libby> it's only recommended


* Using something like homepage as an InverseFunctionalProperty

16:32:46 <danbri> unique identifiers for events are really really
slippery... so many different sense of event
16:33:05 <danbri> event in the social sense of 'meeting' is maybe
easier, as they can have homepages


We didn't make any decision, but I thought it might be useful to raise
the issue here.

Any thoughts about the best way forward?

Maybe we don't want to give the event an ID explicitly, as danbri
suggests: maybe foaf:homepage or somesuch would be ok, although many
eveents will not have a homepage.

My inclination is that some inverseFunctional property would be good as
there are likely to be many ids for the same event, as described by
different people.

Alternatively, we could use owl:sameAs instead if we gave the event
itself a uri, as Nick Gibbins pointed out in a different context:

http://rdfweb.org/pipermail/rdfweb-dev/2003-August/011829.html

An additional issue is what sort of identifier we want if we want
one - go with UUID in the form proposed in iCalendar RFC 2445, or
another form; or some URI, or what.

cheers

Libby

Received on Tuesday, 19 August 2003 10:46:26 UTC